Jump to content

A K-402-Based Full-Range Multiple-Entry Horn


Chris A

Recommended Posts

13 hours ago, Chris A said:

The rest of the story on thicker horn material is just good engineering practice.

 

Three ways to handle an enclosure (or horn wall) resonance:

  1. Make it so stiff that the resonant frequency is moved above the frequency range of the driver.
  2. Make it so compliant that the resonant frequency is moved below the frequency range of the driver.
  3. Make it so massive that it basically doesn't move at all.
  • Like 1
Link to comment
Share on other sites

Door #3 whenever I can do it is the ideal solution, especially with horns.  I think that was PWK's ideal solution, too.

 

Door #4: make the resonant frequency higher via a stiffer structure, then damp the mouth of the horn until there is no measurable resonance via clamping it to a piece of wood or MDF, etc.

 

I like door #4 as an engineering solution to the problem, and I believe that's what PWK actually did with his loudspeakers and horns.  That's also the way that I intend to do it with the future MEH using the thicker/stiffer horn assembly, and clamping the horn mouth to the box.  But that doesn't relieve the designer of having to make the horn walls thick enough not to deflect measurably (especially acoustically significant deflections).  That means that the walls of the horn can't go into "flat plate breakup modes", instead of the "bell ringing" modes that are typical of metal horns--like the early K-400-series horns.

 

I think that a lot of people were stuck with multicell (sectoral) horns from Altec, et al. that couldn't be clamped to a big piece of wood, etc. so they started to put mastic on the back of the horn in the hopes of achieving door #3 solution.  That approach on curved mouth horns doesn't usually work very well--in my experience.

 

468939466dacff0d96b60df3ea4ca3b3.jpg

 

Chris

Link to comment
Share on other sites

  • 3 weeks later...

Today I used a completely different Xilica settings setup schema to parallel the type of crossover design found in the Danley SH-90, which basically consists of using first order crossover filters.  There are no low pass filters used on the woofer or midrange channels in the SH-50.  So to parallel the SH-50 crossover schema in the K-402-MEH, I used first order filters with the MEH (which I'd been using since last fall), but I added the embellishment of the following:

 

1) raised the woofer/midrange crossover point to 550 Hz (from 425 Hz)

2) removed the woofer low pass filter and reset the relative delays

3) removed the midrange low pass filter (was at 6.2 kHz) and reset the relative delays

4) completely re-EQed all channels from scratch and reset the channel relative gains

 

In the end I re-inserted the first order low pass filter on the woofer channel because I found that I needed 4" of acoustic delay to compensate for the woofer ports being in front of the HF compression driver port (on the horn apex).  The relative performance plots follow.

 

First, spectrograms, with the full high pass/low pass filter set used since last fall:

2132611835_K-402-MEHHPandLPfilterSpectrogram.thumb.jpg.89bcdbd39e1439f6feaff45977f58f91.jpg

 

and the new Danley-style (no LP filter on the midrange channel)

 

1773910477_K-402-MEHHPbutnoLPmidrangefilterSpectrogram.thumb.jpg.52bf34fe5d8ee9d0128486ded2b378b4.jpg

 

Next, we can see the difference in the step response, with the newer Danley-style setup cleaning up the pre-rise in the step before 0 ms:

1457462969_K-402-MEHStepResponses(redmidrangeLPgrnnomidrangeLP).thumb.jpg.2be4c5bc8c924f572ebb95a1a36ef288.jpg

 

and the relative phase responses showing the flatter phase through the two crossover regions of the Danley-style crossover (green trace):

 

704717305_K-402-MEHPhaseResponses(redmidrangeLPfiltergrnnoMRLPfilter).thumb.jpg.39927f32f9038d2b6ca69fb4aac76905.jpg

 

and finally, the group delay plots (same colors) showing the lower group delay in the woofer-midrange crossover region and the absence of a midrange-tweeter disruption (basically a wash in terms of group delay):

 

2131115444_K-402-MEHGroupDelayResponses(redmidrangeLPfiltergrnnoMRLPfilter).thumb.jpg.c7e708692220ad5f4f76171c62dc6125.jpg

 

Chris

Link to comment
Share on other sites

To compare phase response of the SH-50 (with corrective EQ) in orange trace vs. the latest K-402-MEH using the Danley-style crossover filters (i.e., no midrange LP filter) in green trace:

 

547824753_SH-50(wcorrEQ)vs.K-402-MEHphaseresponse.thumb.jpg.cb7358382bcc7879474c9f988c619f50.jpg

Link to comment
Share on other sites

39 minutes ago, ellisr63 said:

...Would it have been better if the ports were in a different spot?

No, not really.  The MEH design relies on the phase lag of the electrical crossover filters to time align the drivers.  I just independently confirmed that fact today.

 

35 minutes ago, ellisr63 said:

What would have happened if the woofers had been mounted x" away from the horn sides...would this have given you the acoustic delay you needed? 

Strangely, no it really wouldn't add the needed time delay (unless they are 4" long and narrow ports) but the added air compliance of spacing the woofers farther away would compromise the HF performance of the woofers both in amplitude and phase. 

 

Actually, adding electrical crossover filter delay is not an issue. In fact it says that the electrical crossover network can work to your advantage in the design, and having the crossover simplifies the equalization of the loudspeaker. 

 

The phase growth of MEHs, like these two designs exhibit, has about half the total phase growth as is found in even the most expensive direct radiator designs--which usually have a minimum of 720 degrees of phase growth from 20 kHz down to their bass -3 dB point.  These two designs have significantly less phase growth.  This is something that not a lot of people realize: the performance of MEHs from an amplitude, phase, group delay, time alignment, power response (consistent coverage) and modulation/compression distortion standpoint is probably the best performance that one can get. (YMMV.)

 

Chris

  • Like 2
Link to comment
Share on other sites

Signal going to a driver which is bandwidth-limited at its upper end (woofer, squawker) goes through an inductor (coil) which causes a delay in the current waveform of 45 degrees.  If that waveform took 1 second to make a complete cycle, after the inductor it would start 1/8 second late, then end also 1/8 second late.

  • Like 1
Link to comment
Share on other sites

2 hours ago, parlophone1 said:

Chris, would you explain how can electrical crossover network work to an advantage and simplifies the equalization of the speaker?

I think got a few of your explanations above, but can not quite put it all together.

The SH-50 woofers don't roll off on the high end smoothly.  They have a response peak up around 1.5 kHz and a much smaller one at 2.6 kHz that interfere with the smooth roll-off, and which affect the combined response of the woofers, midranges, and tweeter.  If the woofer response is rolled off at 6 dB/octave using the electrical filter, it puts those 1.6 and 2.6 kHz response peaks much further down in SPL, thus simplifying the EQing of the addition of all three channels--which are all using notch filters in the crossover network individually to flatten the frequency response.  Without the low pass filter on the woofer channel, all three channels would be significantly contributing to the ~700-3000 Hz band--which is actually too complicated to EQ easily.  The woofer channel low pass roll off filter also pushes the time delay of the woofers back into time alignment with the midranges.  Two for one.

 

One thing that few people understand: there is no reason not to let the drivers in an MEH significantly overlap their frequency response bands, since their outputs sum within the aperture of the horn and their polars are also enforced by the horn to be seamlessly summed in SPL and directivity, i.e., there is no lobing.  This makes channel overlap a good thing--especially if the natural HF roll-off of the woofers and midranges is sort of abrupt (like the K-55/K-400 midrange response is in a Khorn or La Scala, which for many year models also don't have a low pass electrical filter). This is probably the first major revelation of the MEH design that became apparent with the initial setup of the K-402-MEH crossover settings.  The rules that work for separate-horn or direct radiator flat baffle loudspeakers don't apply to MEHs in this regard.

 

Chris

  • Like 1
Link to comment
Share on other sites

One of the most important findings of this exercise is that there is no time delay needed for the tweeter diaphragm of the BMS 4592ND (fully broken in) relative to the midrange diaphragm.  This means that tri-amping is actually not required, and all that is needed is a simple cap (~2.5 mF) to block the tweeter channel to provide high pass response and a resistor on the midrange diaphragm channel to provide an additional 5 dB of attenuation (or you can add attenuating PEQs on the HF driver channel to pull down the midrange diaphragm response another 5 dB relative to the tweeter diaphragm).

 

EQing the dual diaphragm driver to have flat response is the only issue.  The following is what I dialed-in for the midrange diaphragm channel (EDIT: 19 Dec 2019 to reflect the newer fractional order crossover filters):

 

1787047433_BMS4592NDMFchannelscreenshot(fractionalorderxover).thumb.JPG.b060fe4eb54d8058676a015e268e4e82.JPG

 

and the tweeter diaphragm channel:

607495437_BMS4592NDHFchannelscreenshot(fractionalorderxover).thumb.JPG.d84249312810f5989754e00d161c3a64.JPG

 

Here are the input channel PEQs across the crossover interference band:

1996443394_BMS4592NDINPUTchannelscreenshot(fractionalorderxover).thumb.JPG.1dbdde46b29d0f14ac7353ae84983542.JPG

 

Chris

  • Like 2
Link to comment
Share on other sites

13 hours ago, ellisr63 said:

So how does it sound now?

As I mentioned in the Subjective Auditory Effects of Quasi-Linear Phase Loudspeakers thread, it sounds more "solid", but not necessarily better.  I attribute that to having more overlap on the 4592 ND midrange/tweeter diaphragm output, however, I can't see any differences in the spectrogram plot, i.e., time vs. frequency vs. SPL, in terms of ringing or interference since the frequencies are so high and the phase relationships so hard to see. 

 

I will do an A-B listening test soon of the two phase/PEQ configurations, perhaps today.  I'm not sure that I'll be able to say how the sound differs, but I will likely be able to say which one I like more.  I'll be listening for detail on the cymbal transients, etc. among other HF reproduction artifacts.

 

Chris

  • Like 2
Link to comment
Share on other sites

On ‎3‎/‎13‎/‎2019 at 6:31 AM, Chris A said:

but I will likely be able to say which one I like more.

 

The opinions we read on the Klipsch forum sometimes influence our perspectives about loudspeakers, so I do appreciate that you show and explain the reasons you arrive at your decisions.

Link to comment
Share on other sites

Well, after looking through my disc collection for multichannel recordings looking for one that has a lot of center channel dynamics and high frequencies like cymbals and trumpets, I used a multichannel recording of Gordon Goodwin's Big Phat Band Swingin' For the Fences (which is one of perhaps 4-5 reference discs that I use for these purposes). 

 

I've found that the Danley-style crossover without low pass on the midrange channel sounds definitely cleaner and clearer, much less harsh (and remember that we're talking about the loudspeakers dialed-in to within ±1 dB from flat response at those frequencies), and more immediate and natural sounding, yet again.

 

So if I were to put my finger on what the Danley SH-50 is doing with its previously "mysterious" crossover design...now a bit less mysterious...you're reading about it here and in the Subjective Auditory Effects of Quasi-Linear Phase Loudspeakers thread, reproduced here for convenience's sake:

 

Quote

When you look at the Danley SH-50 crossovers, you have a 6 dB/octave high-pass slope on the midrange, but no apparent filters at all on the midrange low pass, plus there is attenuation.  The woofer looks to have only notch filtering at ~160 Hz and a low pass of first order above 250 Hz.  The tweeter has a really deep-and-wide notch filter centered at 2.68 kHz [EDIT: actually three big notch filters: one each at 1.4, 2.8 and 6.1 kHz with attenuating gains of -15.75 dB, -10.25 dB, and -8.25 dB, respectively...and bandwidths of about 0.5 octave] but no apparent high pass filter evident in the frequency response.  All three channels have fusing on board their extensive crossover boards.  The combined response of the three channels is basically "a work of art" (i.e., not a clean monotonically decreasing downward slope between driver channels).  I'm not really sure why Danley did it that way--but I suspect that arrayability of loudspeakers played a disproportionate role, as well as a requirement for extremely flat phase response and super high SPL output capability.  Those are not home hi-fi requirements-except for the phase flatness one...in retrospect. 

 

The SH-50 is clearly designed around a "phase link" concept, i.e., a "bridging midrange" between tweeter and woofer in order to linearize the resulting loudspeaker phase response.  See the following link for a definition of "phase link" design (warning: there's some math here): http://www.tonmeister.ca/wordpress/2015/10/29/bo-tech-uni-phase-loudspeakers/

 

It's actually fairly shocking how little frequency response that the four midranges provide in the SH-50...about one octave only, and the resulting frequency response contribution of the midranges is decidedly not flat within that octave.

 

So I think that I'll accept the higher overall phase growth than what I was using beforehand, but lower overall group delay of using the settings which omit the midrange low pass filter and that eliminate the pre-rise in the impulse response that I had.  The difference is audible...and not subtle.

 

Chris

  • Like 1
Link to comment
Share on other sites

By the way, here's an SPL and phase plot of the K-402-MEH after my latest settings update yesterday.  Note the -3 dB point on the bass bin (vertical divisions are 1 dB) which was achieved without ramping filter on the bass bin and with a simple +5 PEQ at 25 Hz, 0.26 octave (Q=5.5)... :biggrin2:

 

1295535114_K-402-MEHwithDanley-StyleCrossoverFiltersand25HzBassBoost.thumb.jpg.91a48d835ee391d841113cd3718e9408.jpg

 

The MEH is positioned touching the front wall at mid-wall height in the center of the room, with the microphone on-axis at 1 metre distance.  It sounds amazing playing multichannel organ discs and movies with deep LF effects

 

Chris

  • Like 1
Link to comment
Share on other sites

I was talking about the two 15" woofers in the box/horn, unported.  The harmonic distortion levels seem to be much lower than the TH subs. Routing movie LFE to the subs--away from the rest of the music LF content going to the MEH--seems to work extremely well.

 

I spent some time with REW and microphone + Xilica to carefully dial-in each channel of the MEH, including the two BMS 4592ND channels, with the result you see plotted. Getting the phase right seems to be the most time consuming part and EQing each channel to flat SPL within its respective frequency band is key to that task, then combining the channels with proper relative gain (and relative delays) makes the task much more straightforward.

 

Note that if the MEH were in the room corner, it picks up something like 12 dB of added response at 40 Hz, and so it would be EQed for that particular boundary loading.  The measurement above is loaded at mid-wall, elevated ~46" (to centerline) above the floor.  I had been planning on an integrated subwoofer that would serve as a riser/base, but it looks like that may not be needed.  More listening is on the agenda.

 

Chris

Link to comment
Share on other sites

A question on integration of the MEH into your system as a center Chris.  How do you run 2 channel playback through your Onkyo pre/pro?  Direct with no center?  Some kind of synthesized center like "3 channel stereo"?  Do you find yourself processing 2 channel recordings into surrounds and sub as well?  I don't quite have the balls to cut one of my K402s yet but I'll be experimenting with one as a center on a Cornwall bass bin driven by a 4592ND this weekend - so just wondering how you get the benefit with 2 channel recordings.  

Link to comment
Share on other sites

3 hours ago, Delicious2 said:

How do you run 2 channel playback through your Onkyo pre/pro?  Direct with no center?  Some kind of synthesized center like "3 channel stereo"?  Do you find yourself processing 2 channel recordings into surrounds and sub as well?

I generally find that the "Stereo" mode works best, effectively pulling in each TH sub behind each Jub to make the Jubs into three-way loudspeakers--although the Jub bass bins do play in parallel to the subs.  It most often sounds better in Stereo mode than the synthesized modes on the AVP.  If talking about music, "Direct" or "Pure" discard the subs if an embedded subwoofer channel is not present, and I find this doesn't sound nearly as good.  I run my TH subs 100% of the time.

 

For movies in stereo--I find the opposite to be true (Neural THX and THX Cinema modes seem to perform better consistently).  There are a lot of British/Australian/Kiwi movies and streamed TV shows that are in stereo, so those synthesized modes put the dialogue on the center channel (where it increases clarity and speech recognition substantially). 

 

There are of course many synthesized modes available on the AVP, so an exhaustive discussion of source material vs. listening mode would be pretty long.  I play around with the different modes from time to time and find one or more that work better than stereo for movie soundtracks.  Very rarely, stereo music can sometimes sound better on a synthesized mode, but I find that some sort of Dolby or perhaps DTS embedded encoding is usually found in the stereo music tracks for them to sound better.

 

3 hours ago, Delicious2 said:

I don't quite have the balls to cut one of my K402s yet but I'll be experimenting with one as a center on a Cornwall bass bin driven by a 4592ND this weekend - so just wondering how you get the benefit with 2 channel recordings.  

My multichannel music library is at around 200 discs and growing.  This group of discs holds virtually all of my reference recordings that I use for system fine tuning, troubleshooting, and listening pleasure.  If good multichannel recordings were available for all music genres and artists that I listen to, my multichannel library would be much larger.  I'm beginning to devote an increasing fraction of my disc buying budget to multichannel. 

 

Stereo recordings just sound much thinner and without significant depth, presence, and sense of envelopment by comparison (usually like the difference between the living and the dead)...although most people on the forums don't want to talk about that subject--so I typically don't :wink:

 

Chris

  • Like 1
Link to comment
Share on other sites

1 hour ago, Chris A said:

I generally find that the "Stereo" mode works best, effectively pulling in each TH sub behind each Jub to make the Jubs into three-way loudspeakers--although the Jub bass bins do play in parallel to the subs.  It most often sounds better in Stereo mode than the synthesized modes on the AVP.  If talking about music, "Direct" or "Pure" discard the subs if an embedded subwoofer channel is not present, and I find this doesn't sound nearly as good.  I run my TH subs 100% of the time.

Well, that kind of answers it - you use stereo mode to pull in the subs.  I get that you don't listen to much 2 channel music because it just isn't as satisfying compared to multichannel but what about the MEH with 2 channel music to anchor the image in the center?   some modern version of PWKs center channel circuit?  something derived from the 2 channels but not synthesized in the sense of adding any effects.  Maybe just another memory preset on the Xilica could create this when there is no actual center channel.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...